The neural network for grouping wells of different facies zones of productive layer

International Conference on Innovations and Prospects of Development of Mining Machinery and Electrical Engineering 2018

矿业工程;电工学

Chudinova, D. Yu.^1,2 ; Kotenev, Yu. A.^1,2 ; Sultanov, Sh. Kh.^1,2 ; Mukhametshin, V. Sh.^1

Ufa State Petroleum Technological University, Kosmonavtov St., 1, Ufa, Bashkortostan

450062, Russia^1

State Autonomous Scientific Institution Institute, 129/3, October avenue, Ufa, Bashkortostan

450075, Russia^2

关键词: Four-group;    Group recommendations;    Parameters characterizing;    Spatial relations;    Technological parameters;    Training sets;    Western siberia;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/194/8/082008/pdf
DOI  :  10.1088/1755-1315/194/8/082008

来源: IOP

【 摘 要 】

Grouping of many features of a well stock of layer of the large-scale deposit of oil of Western Siberia with use of artificial neural network was carried out. For grouping the initial set including 555 objects was used, 95 % were chosen from them as the training set and 5 % as test. For training of neural network 17 features characterizing both geological and physical, and technological parameters of layer were accepted. Based on the results of tuning and subsequent training of the neural network, four groups of wells were identified, the closest in their geological and technological parameters. For each group of wells, in operation, parameters characterizing the uniqueness of the selected group were described. The binding is given to localization in the spatial relation of layer and to remaining reserves of oil. For each group recommendations about involvement of remaining reserves of oil in active development were offered.